Implementation of a half-precision floating point number in .net standard
Half is not fast enough and precision is also very bad, so is should not be used for mathematical computation (use
Single instead). The main advantage of Half type is lower memory cost: two bytes per number. Half is typically used in graphical applications.
- Fast Half Float Conversions, Jeroen van der Zijp, link: http://www.fox-toolkit.org/ftp/fasthalffloatconversion.pdf
- IEEE 754 revision, link: http://grouper.ieee.org/groups/754/